## Math #### Matlab Transfer Function + Laplace ``` s = tf('s'); % define laplace variable sysG = (3*s + 7)/(2*s^2+5*s+6) % transfer function ``` ### LTI System - **SISO**: Scalar Input, Scalar Output - **Causal:** Output depends only on input up to that time - **Linear** - If $u(t)\to y(t)$, then $qu(t)\to qy(t)$ - If $u_{1}(t)\to y_{1}(t)$ and $u_{2}(t)\to y_{2}(t)$, then $u_{1}(t)+u_{2}(t) \to y_{1}(t)+y_{2}(t)$ - **Time Invariant** - Shifting input in time shifts output correspondingly ### Unit Step $ \underline{1}(t)=\begin{cases} 0 & \text{if }t<0 \\ 1 & \text{if } t\geq0 \end{cases} $ $ \underline{1}(t)= \int_{-\epsilon}^{\infty} \delta(t) \, dt $ #### Step Response ``` t = 0:dt:t_max % time vector y = step(sysG,t); % step response plot(t,y); % plot step response ``` ### Unit Impulse $ \delta(t) = \begin{cases} 0 & \text{if }t\neq0 \\ \infty & \text{if } t=0 \end{cases} $ $ \delta(t) = \frac{d}{dt} \underline{1}(t) $ ### Impulse Response Function Response of the system to the input $u(t) = \delta(t)$ - Complete description of the system’s response to any input $u(t)$ given zero initial conditions - $h(t)<0=0$ due to causality - Integrate starting from $- \epsilon$, an infinitesimally small number ### Euler’s Equation $ e^{\pm j \omega t} = \cos(\omega t) \pm j\sin(\omega t) $ ### Convolution Integrates the impulse response function over any input $ y(t) = \int_{-\epsilon}^{t}u(\tau)h(t-\tau)~ d\tau $ ### Fourier Transform Convert a function from time domain to frequency domain $ \int_{-\infty}^{\infty} f(t)e^{-j\omega t} \, dt \to F(\omega) $ Doesn’t work when $f(t)$ doesn’t go to zero as $t\to \pm \infty$ [3b1b video](https://www.youtube.com/watch?v=spUNpyF58BY) ### Inverse Fourier Transform $ f(t) = \int_{-\infty}^{\infty} F(\omega)e^{j\omega t} \, d\omega $ ## System Modeling | | Mechanical | Electrical | Fluidic | Thermal | | -------------------- | ------------------ | ------------------------- | ------------------------------------------ | ------------------------------------ | | Force / Voltage | $F$, [N] | $V$, [V] | $\Delta P$ [Pa] | $\Delta T$ [K] | | Velocity / Current | $v$, [m/s] | $I$, [A] | $Q$, [$\text{m}^{3}/\text{s}$] | $Q$, [J/s] | | Inertia / Inductance | ${} F=m\dot{v} {}$ | $V=L \frac{dI}{dt}$ | ${} \Delta P = \frac{\rho L}{A}\dot{Q} {}$ | N/A | | Damping / Resistance | ${} F=bv {}$ | ${} V=IR {}$ | ${} \Delta P=QR_{f} {}$ | $R = \frac{L}{kA}$ or $\frac{1}{hA}$ | | Spring / Capacitance | $F=kx$ | ${} I=C \frac{dV}{dt} {}$ | $Q = \frac{A}{\rho g}\Delta \dot{P}$ | $Q=mc\Delta T$ | ## Laplace Transform Convert a function from time domain to s-domain $ L[f(t)] = \int_{-\epsilon}^{\infty}e^{-st}f(t) \,dt = F(s) $ $s$ is a complex number which is a function of $\omega$: $ s = \sigma + i\omega $ $\sigma$ is chosen to make the transform converge ([steve brunton explanation](https://youtu.be/7UvtU75NXTg)) $\epsilon$ is a infinitesimally small number that ensures the bottom bound of the integral is zero ### Laplace Identities ![[Pasted image 20241007002201.png|400]] ### Common Laplace Transforms ![[Pasted image 20241007002227.png|400]] ![[Pasted image 20241009004708.png|400]] $\begin{array}{l} \\ F(s) & & & f(t) \\ \frac{2\omega s}{(s^{2}+\omega^{2})^{2}}& & & t\sin\omega t \\ \frac{1}{(s+\alpha)^{2}} & & & te^{-\alpha t}\\ \frac{\omega}{s^{2}+\omega^{2}} & & & \sin\omega t\\ \frac{s}{s^{2}+\omega^{2}} & & & \cos\omega t\\ \frac{s+a}{(s+a)^{2}+b^{2}} & & & e^{-at}\cos bt \\ \frac{b}{(s+a)^{2}+b^{2}} & & & e^{-at}\sin bt \\ \frac{1}{s+a} & & & e^{-at} \\ \frac{1}{(s+a)^{2}} & & & te^{-at} \end{array}$ ### Harmonic Signals Replace sinusoidal signal with difference of complex exponentials: $ \sin(\omega t) = \frac{e^{i\omega t}-e^{-i\omega t}}{2i} $ ### Derivatives $ L(\dot{f}) = sF(s) - f(-\epsilon) $ (derivation via integration by parts) Formulas exist for higher derivatives such as: $ L(\ddot{f})=s^{2}F(s) - sf(-\epsilon) - \dot{f}(-\epsilon) $ or more generally, $ L(f^{m}(t))=s ^{m}F(s) - s ^{m-1}f(-\epsilon)-s ^{m-2}\dot{f}(-\epsilon)-\dots-f^{(m-1)}(-\epsilon) $ If zero initial conditions, this generalizes to $ L(\left( \frac{d}{dt} \right) ^{n}f~) = s^{n}F(s) $ ### Integrals Similarly, by integration of parts, we have $ L\left( \int_{-\epsilon}^{t} f(\tau) \, d\tau \right) = \frac{1}{s}F(s) $ and nested integrals increase the negative power of s: $ L\left( \iint f \right) = \frac{1}{s^{2}}F(s) $ ## Transfer Function Convolution in the time domain is equivalent to multiplication in the $s$ domain $ Y(s) = H(s)U(s) $ $H(s)$ is a complete description of an LTI system given zero initial conditions ## Inverse Laplace Transforms If $s$-domain gives us a sum of pure harmonics or exponentials, we can simply look up the inverse Laplace transforms for each part. ### Cover-Up Method Taking the inverse of rational functions with partial fractions. The function $ F(s) = K \frac{(s-z_{1})(s-z_{2})\dots(s-z_{m})}{(s-p_{1})(s-p_{2})\dots(s-p_{m})} $ can be rewritten in the form $ F(s) = \frac{C_{1}}{s-p_{1}} + \frac{C_{2}}{s-p_{2}} + \dots + \frac{C_{n}}{s-p_{n}} $ For each $i$, multiply both sides by ${} s-p_{i} {}$ and let $s=p_{i}$ such that we are left with only $ C_{i} = (s-p_{i})F(s) $ Plugging in the coefficients and taking the inverse Laplace transform gives us $ f(t) = \sum_{i=1}^{n}C_{i}e^{p_{i}t} $ ### Matlab Partial Fraction Expansion ``` num = conv([1 z1], [1 z2]); % numerator den = conv([1 p1], [1 p2]); % denominator [r,p,k] = residue(num,den); % residues ``` resulting in `r = [C1 C2 C3]` and `p = [p1 p2 p3]` ## Diff Eqs ### Final Value Theorem If all poles of $sF(s)$ are in the left half of the $s$-plane, then $ \lim_{ t \to \infty} f(t) = sF(0^{+}) $ ### Initial Value Theorem $ f(0^{+}) = \lim_{ s \to \infty } sF(s) $ ### Solving ODEs $ m\ddot{y}(t) + by(t) + k = u(t) $ 1. Take Laplace transform of both sides and make sure to account for initial conditions: $ \begin{align} &L(\dot{f}) = sF(s) - f(0^{-}) \\ &L(\ddot{f})=s^{2}F(s) - sf(0^{-}) - \dot{f}(0^{-}) \end{align} $ 2. Solve for $Y(s)$ 3. Partial fraction expansion via cover-up method or Matlab `residue` function 4. Take inverse Laplace transform of fraction components